We recently launched our new article comments system and had an interesting weekend discussion about AI tools and whether they add value to our lives. The feedback saw a majority sharing the positive impacts that the various tools had in their lives.
However, there are critical issues abound. Former OpenAI founder Andrej Karpathy recently shared an interesting perspective, highlighting how AI could impact the web and its content (via The Decoder).
It’s 2025 and most content is still written for humans instead of LLMs. 99.9% of attention is about to be LLM attention, not human attention.E.g. 99% of libraries still have docs that basically render to some pretty .html static pages assuming a human will click through them.…March 12, 2025
With the rapid emergence of generative AI, professionals from various fields have raised concerns, citing the worries and woes of AI taking over their jobs.
While Karpathy foresees a future where most of the content will be written for humans, the AI guru claims 99.9% of the content will be curated in a way that it would be more friendly to large language models (LLMs).
Still, the way AI consumes information is miles apart from the way humans do. Human consumption requires a coherent flow of thoughts, well-spaced and grouped paragraphs, and more.
AI doesn’t consider elements such as font size and paragraph indention. Instead, it relies on structured data with minimal ambiguity because it ingests absolutely everything, unlike humans, who mostly skim through content to the part that interests them most.
Medium recently highlighted a plausible solution for the issue, citing a dual-publishing strategy — meaning publishers and authors will have to create content for humans and AI.
Both parties have a different view and approach to interacting and understanding information. Medium explained how content created for humans will feature explanations, storytelling, and examples, while content created for AI will feature structured data, tagging, and reasoning annotations.
Related: Bill Gates says AI will replace humans for everything
Finally, content could feature Metadata layers, which would allow AI to interact with the information appropriately without impacting the flow for humans.
Rethinking how we create content for AI
AI is increasingly becoming popular among humans, gaining broad adoption worldwide. As you may know, multiple reports recently surfaced indicating that OpenAI, Anthropic, and Google are reportedly facing challenges in developing advanced AI models because of a lack of “high-quality content” for model training.
However, former Google CEO Eric Schmidt dismissed the claims, indicating there is no evidence scaling laws have begun to stop AI progression. “There is no wall,” added OpenAI CEO Sam Altman.
Elsewhere, multiple publications have lodged copyright infringement complaints against Microsoft and OpenAI. However, CEO Sam Altman admitted that developing tools like ChatGPT and Copilot without copyrighted content is impossible.
There’s seemingly a gray area in copyright law, making it difficult to establish whether it is strictly prohibited to train AI tools using copyrighted content. But as it stands, AI needs human-written content for training and advances.
It will be interesting to see if publishers and authors will embrace new writing techniques to factor in AI as part of their target audience.
